The Control of Motion and Aeration in Corrosion Tests. - Industrial

The Control of Motion and Aeration in Corrosion Tests. J. F. Thompson, and R. J. Mckay. Ind. Eng. Chem. , 1923, 15 (11), pp 1114–1118. DOI: 10.1021/...
1 downloads 0 Views 941KB Size
1114

INDUSTRIAL A N D ENGINEERING CHEMISTRY

ards or by an accurate determination on a long tube, Calibrated viscometer. ACKNOWLEDGMENT The writer wishes t o express his indebtedness to J. R. Powell, chief chemist of the Armour Glue Works Laboratory,

Vol. 15, No. 11

under whose supervision this investigation was carried out, for his many helpful suggestions and cooperation; to Robert H. Bogue, of Lafayette College, for his aid in the calibration of the MacMichael instrument, and to W. H. Herschel, of the Bureau of Standards, for valuable suggestions and kind review of the paper.

The Control of Motion and Aeration in Corrosion Tests’ By J. F. Thompson and R. J McKay THEINTERNATIONAL NICKEL Co., NEWYORK, N. Y.

Exact results can be obtained in corrosion tests by liquid immersion CID corrosion tests The results obtained in when quantitative control of the aeration of the corroding liquid of several “acid-rethese tests (Table I) showed and the motion of the corroding liquid relative to the corsisting” metals were such R surprising increase in roded solid is maintained. Tests made to predict the practical undertaken a t the Mellon the rate of corrosion and in service of factory equipment will be in error unless these important Institute of Industrial Rethe accuracy of results a t factors are duplicated, or accurafely corrected for. Control is more search during 1920to detereven low velocities over the dificult in quiet than in moving tests.2 mine the causes of irregurate and accuracy obtained larities in the length of serThe mechanism by which variation in aeration and rate of motion in quiet tests, that the plan a$ects the corrosion rate is discussed. A test method of control vice obtained from Monel of making quiet tests was suficiently accurate to reproduce results within about 5 per cent is metal apparatus used for discarded as useless. They given. Also, results obtained in connection with a series of 2000 pickling steel sheets. As could not furnish a predictests on acid-resisting metals in 2 to 10 per cent sulfuric acid, duthe tests progressed the tion of practical results, nor plicating some conditions found in pickling steel sheets, are precontrol of certain factors were they accurate enough sented. The results illustrate an increase in rate of 500 per cent was found necessary to obfor any conclusions. from air-free to air-saturated solution and of 600 per cent from tain accurate results, but in No increase in the rate comparative quiet to motion at 0.5 foot per second, but the agreement spite of the importance of was apparent between 0.25 has been well within 5 per cent with quantitafive control of these these factors a search of and 0.5 foot per second. oariables. The results are in accord with experience in pickling the literature revealed little The percentage experimenpractice. or no accurate information tal error here is less than regarding them. was usually obtained, but A large part of the literature on related subjects discusses the usual effect of a much layger error in quiet than in tests made by suspending a sample in a solution, making moving tests is found. Of all tests made the average varino provision for movement of the solution nor for the pre- ation was 4.5 per cent for moving tests and 20 per cent for vention of motion. It seemed from the outset that the quiet tests. duplication of results could not be obtained by this method, TABLE I as it could not compensate for the change in concentration Air saturation about 15 per cent. Temperature 60° C. 6 oer cent sulfuric acid of the corroding agent, which must of necessity take place a t Motion -Corrosion Rate-Mg./Sq. Dm./DayAverage the solid-liquid contact. Therefore, an apparatus was Rate Sample Sample Sample Variations Ft./Sec. 2649 2661 2664 Average Per cent planned which would give a positive regulation and control 22 24 0.0 26 23 7 of the relative motion of the sample and the solution, with0.125 141 138 139 139 0.7 0.50 140 130 135 135 2.3 out any very definite idea of just how this movement would a Calculated by averaoing the percentage variations of single tests affect the mechanism and the rate of corrosion. One of the from their average. This Gas found to be a useful measure of accuracy. most generally used machines in pickling sheets produces a motion in the solution of a character suitable for laboratory As the tests had for their purpose the study of corrosion of control, and by copying the motion of this machine it was metals in the pickling of steel, they were made under condipossible to regulate the movement and a t the same time to tions as nearly as possible duplicating those of actual picksubject the samples to the same conditions in this respect ling practice. The temperatures used ranged from 60” to that obtain in actual service. A laboratory apparatus 90” C., and the concentrations of acid from 2 to 10 per cent duplicating this motion was therefore devised. of sulfuric acid by weight. EFFECTOF MOTIONON RESULTS APPARATUS FOR MOTION CONTROL

A

After the apparatus was installed, tests were run to determine the difference in the effect of motion or quiet, both from the standpoint of absolute corrosion rate and of reproducibility of results. The quiet and the moving2 tests were made in all respects under the same conditions, except that in the quiet tests the drive motor of the moving mechanism was shut off. Received July 7, 1923. The terms “quiet” and “moving” are used in preference to “static” and “dynamic,” favored b y other writers, because the latter properly are used in reference to forces rather than to matdrials: 1

2

The movement of the samples relative to the solution was accomplished by the apparatus shown in Fig. 1. Virtual, harmonic, vertical motion was transmitted to a horizontal bar from an electric motor through a belt drive with cone pulleys, worm reducing gear, chain and cone sprockets, and adjustable crank. The motor was 0.5 horse power, 1700 revolutions per minute. The length of vertical stroke of the bar was adjustable in the yoke from a minimum of 0.5 inch to a maximum of 4 inches. The rate of stroke could be adjusted from 3 to 100 revolutions per minute. These adjust1

I N D UXTRIAL A N D ENGINEERING CHEiMIXTRY

November, 1923

.

I

ments enabled a variation in the rate of movement of the sample through the solution from 3 to 800 inches per second. Glass rods, bent in such a shape as to carry the sample with contact a t only two places, passed through holes in the bar and were fastened rigidly by thumbscrews. Holes were provided for five rods, and, in cases where no inaccuracies could result from testing more than one sample in the same solution, a frame was used to carry a number of rods. This device furnished a regular and measurable movement, easily adjusted. The samples came into contact with the rod a t only two lines, each 0.25 inch in length, and the sample needed no suspension holes or other special shape. The rate of movement was controllable independently of all other factors. Batlery jars with a capacity of about 4 liters were placed in a large water bath on a table under the moving bar, a t such a height that the glass rods could be easily set to carry the samples completely immersed, still allowing freedom of movement and easy removal. The table carrying the samples rested on the floor, independent of the frame which carried the motor and other moving parts, so that there would be no vibration to disturb the contact film. The fact that the motion was noncontinuous was considered a disadvantage, because there was a complete stop and also a maximum rate 50 per cent higher than the average. However, it was considered more satisfactory than a continuous rotation, because the swirling produced by rotation would make the actual velocity past the surface harder to determine. A continuous-flow apparatus to handle the solutions in question a t known speeds, possible of regulation within wide limits, was considered too expensive; but, in view of the results obtained, it seems that such an apparatus would pay for itself in results of general value.

1115

results of Test 1 and Test 2 is typical of the variations that were being obtained a t this stage of the research, but the effect of the greater supply of air near the surface is shown in both. It was then suggested that if the air had such an important effect a convincing qualitative proof would be to direct a small stream of air against a plane surface of metal and thus

RESULTSWITHOUT CONTROL OF AERATION After the apparatus was operating, results were obtained for a time which checked each other closely. Then unaccountable variations began to appear. Results varied as much as 20 per cent from one day to another; for no apparent reason. Finally, it was noticed that when the laboratory windows were left open in windy weather and there was an appreciable draft of air through the room, results were invariably high. The only obvious way in which thls breeze could affect the tests was in blowing the steam away as it arose from the jars and thus coming in contact to a variable extent with the surface of the solutions. No reference could be found in the literature to differences in corrosion rate with aq air contact in acid solutions of such a strength, but a consideration of the .chemical reactions involved showed that a considexable effect might be expected. Therefore, the following test was devised to check the point: Sevoral samples of the ‘same alloy were tested in the same solution, placed a t different distances below the surface, while especial precautions miere taken to keep the solutions quiet. The results were obtained in two typical tests. The distance from the surface as given is the center of the line in which the samples moved. The alloy used was Monel metal and the temperature and concentrations were the same qs in Table I. TABLB I1 Distance from Surface Inches 0.75 2.00 3.25

Test 1 Mg./Sq. Dm./Day 193 173 167

Test 2 Mg./Sq. Dm./Day 152

140 127

If the air contact with the surface affects the rate of corrosion, it would be expected that the samples near the surface would corrode a t a more rapid rate than those a t the lower levels. That this actually happens is shown with surprising definiteness by the results. The difference between the

FIG.I-FRONT VIEW

O F THE

APPARATUS

possibly produce an indentation a t the point of contact. This experiment was tried by connecting the laboratory air line to a water-bottle delivery tube, so arranged under a 6 per cent acid solution that the stream of air was forced a t right angles against the sample. The result was that 0.25-inch plate samples of several acid-resisting metals were completely perforated by the air jet before there was any appreciable thinning in the immediate neighborhood of the holes. Photographs of some representative samples are given in Figs. 2 and 3. These results were interesting in that they offered new enlightenment as to the mechanism of the corrosion and they furnished an explanation of certain of the practical phenomena. However, they demonstraked the necessity of control of the air content of the solution, which introduced new experimental difficulties. NECESSITY OF AERATION CONTROL The resistance of so-called “acid-resisting” metals to acid corrosion is due largely to the fact that they do not replace hydrogen in acids. Taking copper, for example, the reaction CU HzSOi ---t CuSOi Hz (1) will not proceed in the direction indicated. Consideration of the heats of reaction indicates that the reaction

+

+ + 2H2S04

~ C U

0 2

+

2CuSOa

+ 2HzO

(2)

INDUSTRIAL AND ENGINEERING CHEMISTRY

1116

n

Vol. 15, No. 11

v,..

will have a greater tendency to take p l a c c i n fact,, it is exothermic, whereas Equation 1 is endothermic. The evident products of corrosion of copper by sulfuric acid are copper sulfate and water, and it, is therefore probable that the reaction proceeds according to Equation 2. Tlierefore, it would be

The oxygen necessary for these reactions may, of course, be furnished by an oxidizing agent., if one is present, or by catalysts. Tbese catalysts may be compounds, such aa iron sulfates, which are oxidized readily by oxygen and reduced by metal, or finely divided or colloidal material carrying absorbed oxygen, or, in special cases, a conibinatioii of the two, as witli the hydrates of iron.' INTERDEPENDENCE OF

Fn;.Z-Molie~ MBTALPBKFOXATBU BY A m

JET

expcct.od that the rate of reaction would depend on t.hc concentration of Cu, Os, and H2S04. In other words, dissolved oxygcn tihoiild be iis active io producing corrosion as a strong acid. Copper is here used as an example, bat the reasoning will apply to any metal wliich will not direct,ly replace hydrogen in acids. For metals which will replace hydrogen, tlie following reactions take dace:

MOTION AND AERATION

Since the reaction prcsumably only takes place at the surface of t,he metal, there is another condition to consider. Inst,cad of t,he rat,e of react,ion being determined by t,hc concentration in the body of the solution, it, is determined by tlie concentration in actual contact. wii.11 tbe surface. As has bcen seeti, the material whose contact must be considered is not the acid but dissolved oxygen. Tlierefore, a change in the mpply of oxygen to this surface determines a change in the rat,r of mot,ion of t,lie solution and the rate of supply of oxygen to it. Thus, the rate of motion of tlie solution is of importance in supplying oxygen to the sample, and, wit,liiii certain limits, t,he coiioentriltion in tlie contact film will vary with tho motion and tile rate of iiorrosioii will increase with increase in niotion. Tinless there is some continuons motion of the solut,ion, the corrosion products will be unevenly distributed, owing to slight movtments of t,he corroding liquid, and this will produce differencesin the concent,ration of the corroding medium. The differences in concentration .i\,ill set up qmsi-conccntration cellss which acce1erat.e corrosion a,t one point and inhibit it in others. Therefore, the corrosion is umveii and such t.ests do not give duplicate results. A controlled moderate agitation of unvarying speed mill largely prevent these effects and enable tbe reproduct,ion o€ results. The agitated t,est is the more practical, because in most service cases of corrosion of acid-resisting metals tlic corroding mediiim is in movement.

As above, it would be expected t.liat the reaction velocity would be increased by t.lie presence of oxygen, hut in this case both reactions are exot.hcrniicand the metal is corroded with considerable speed in bot,li cases. Tbe resiilts given below are in apeement. with this reasoning, as sliowii by the smaller effect of aeration in accclerat,ing the corrosion of iron than of mctals of low solution pressure. (Table V) According to S~ide11,~ the solrbility of oxygen from air in water at 25" C. is 0.0075 gram per liter, or about 0.00094 normal. In tlie present tests t.he concentrations of acid were from 2 to 10 per cent, or 0.5 to 2.0 normal. The conceniration of acid is thus from 500 t.o 2000 times as great as that of oxygen. At 80" C. this ratio would probably be increased to a maximum of 10,ooO. Thua, at tlie riorinal corrosion rate for copper in 2 to 10 per rent, sulfuric acid, loo0 mg. per sq. din. per day, a sample of area of 0.5 sq. dm. (an ordinary slieet sample 2 inches square) would exhaust the air from a 200-cc. corroding solution, according to Equation I, in about 6 minutes, if the eorrosion proceeded at tlie maxiinnin rate. The concentration would be dccreased 16 pcr cent in 1 minute and the rate F ~ G~.- A L C W I N I Y I Baorrzr: PBRBOLATIID SY Am JSl of corrosion diminished accordingly. This is accompanied by no appreciable change in the concentration of the acid. APPARATCS FOR AERATION CO~~\ROL Therefore, it would be expected that in such a corrosion the In order to produce a h o r n concentration of dissolved rata would depend very much more on the concentration of air in solution than on the concentration of acid. This has oxygen in the corroding xolutions, the apparatus shown in been found to be true in the experiments of the authors. k + % e n d , Trans. A m . Eleclrochcm. Soc., 40. 1 (1922).

* "Solubilities

of Organic and Inorpanic Materials." lS17.

'McKay. IMd.. 41, 201 (1922).

INDUSTRIAL A N D ENGINEERING CHEMISTRY

November, 1923

Fig. 4 was developed. An alundum crucible of medium fineness was fitted carefully with a rubber stopper. The stopper was cut thin to insure a perfect fit without breaking the crucible, and over it was placed a wooden disk held by strong, narrow rubber bands passing vertically around the crucible. The disk and stopper were bored to receive a glass tube connected with the pressure gas supply. This arrangement enabled the holding of sufficient air pressure within the crucible to produce a rapid emission of fine bubbles. In use the stopper was so fitted on a bent glass tube as to lie in the bottom of the battery jar under an inverted glass funnel held on a glass standard. The funnel collected all the bubbles from the crucible and delivered them near the surface of the solution, thus preventing actual contact between the bubbles and the samples. The solution flowed up through the funnel with the bubbles, giving sufficient circulation to assure saturation with gas, but without affecting appreciably the average rate of movement relative to the sample. It was necessary, in making a test, to use great care that this saturation device was operating a t full efficiency. A small crack in the crucible or leak a t the stopper allowing an appreciable proportion of air to flow in large bubbles invariably caused low results. The saturated condition was usually indicated by the formation of secondary minute bubbles in the solution from time to time, and unless this happened the saturating apparatus was not considered efficient. In tests with the solution saturated with air the surface of the liquid was in contact with the laboratory atmosphere. In all others a glass cover was placed over the battery jar and neutral or suitable gas or gas mixtures were used in the saturator. The glass cover was a rectangular plate bored to admit the air tube, siphon, and sample support. Vapor from the warm solution condensed on this cover, sealing any slight openings. Interstices were left sufficient to allow the gas to escape. On the moving sample support a glass tube was fastened with a rubber tube in such a manner that the top end was sealed and the lower end fitted in the hole in the cover so as to seal the opening with a water film and still move freely u p and down and allow the necessary play sideways. Thus, by controlling the relative volumes of gas admitted to the saturator, the concentration of oxygen or any active gas could be controlled independent of all other factors. With apparatus as now developed it was again found possible to duplicate results closely. Some typical results are given in Table 111.

Test

TABLE111-MONEL METAI, Rate of motion. 0.25 foot Der second Concentration Corrosion Average Temperature HaSOa Rate Variation O c. Per cent Mg./Sq. Dm./Day Per cent Methane used i n satuiator

1

82 100

1 2 3 4

82

2

6 3 Nitrogen used i n saturator 10

6 6

2

Av.

43 71 135 152 135 127 137

A i r used i n saturator

1 2 3 4

1 2 3 4

4 5

The much better duplication of results where the saturator was used is of note. The tests without the saturator were made during the first part of the investigation before the importance of the factors in oxygen supply discussed above were realized. The comparatively large variation in these tests resulted in the development of the saturator. TABLEIV-COLD DRAWNCOPPER Rate of motion, 0.25foot per second Concentrated Temperature HaSOa Corrosion Rate c. Per cent Mg./Sq. Dm./Day Methane used in saturator 82 6 64

Test

1 2

100

3 A i r used in saturator 60 6 60 2

1 2

78

1070 1110

The results obtained by using a reducing gas in the saturator were always lower than those wherein nitrogen was used. It is believed that this was because the nitrogen (from commercial tanks) contained a trace of oxygen or a slight amount of air was carried to the solution by drip from the glass cover. With the reducing gas this effect would be minimized. G A INLET ~

-h

H RUBBER STOPPER

PLATEGLASS

G L A 5 5 JAR WATER

LEVEL

SPECIMEN

GLAS3 TUBING

GLA5S FUNNEL

POROUJ CRUCIBLE

FIG.4-AERATING

APPARATUS

Tests made on about fifteen acid-resisting metals and alloys, with and without the presence of air, showed effects of the same order of magnitude in all cases.. According t o the reasoning in this discussion, the effect of oxygen in the corrosion of steel should be less than in the corrosion of Monel metal. This inference was found to be true, and the relative effect on steel was found to be greater in the lower acid concentrations, which observation agrees with the interesting contemporary results of Whitman, Russell, Welling, and Cochrane.6 TABLEV-RATE OF SOLUTION OF COLD ROLLED STEEL Acceleration Sam- Area Weight Ratio Loss Saturated by Air ple Sq. Dm. G. Loss Mg./Sq. Dm./Hr. with Per cent Test 1: 82' C . , 6 pev cent HaSOa, 5 minutes A 0.148 21.7491 0 6851 56 Air 0 B 0.143 18.1341 0 7492 63 Nitroeen Test 2: 82' C . , 0.05 per cent HaSOI, 15 minutes A 0.13 20.0935 0,0395 1.21 Air 22 B 0.12 17.0331 0.0296 0.99 Nitrogen Test 3: 82 a C., 0.05 per cent HaSOa, 5 minutes A 0.13 19.9352 0.0300 0.92 Nitrogen B 0.12 17.1051 0.0341 1.14 Air 24 I

N o saturator.

1 2 3

5.3

1117

Aiv in contact wilh surface of the solution

e THISJOURNAL, 15, 672 (1923).

INDUSTRIAL A N I ) ENGINEERING CHEMISTRY

1118

The many other factors which affect the rate of corrosion of metals will not be discussed, but the inadequacy of drawing conclusions from experiments in which agitation and aeration are incompletely controlled is to be emphasized. The following notes mention briefly some of the other conditions existing throughout the tests. The duration of tests was usually 20 hours. In the absence of serious pitting, the loss in weight per unit surface per unit time was used as the measure of the corrosion rate. Tests were repeated until the certainty of securing duplicate results was assured. The samples were of a standard size, a disk 1 inch in diameter and 0.25 inch thick. All surfaces were ground with zero emery before testing. Solutions were analyzed before and after all tests. A constantlevel siphon furnished water to replace the rapid evaporation. The volume of solution was 3 liters. The sample was set on edge on two glass rods, making contact at only two places. Temperature was maintained by an electrically heated, thermostat-controlled bath covered with paraffi.

Vol. 15, No. 11

This investigation was begun for the purpose of studying certain practical problems and led to their solution before the effect of variation in rate of motion and of concentration of dissolved oxygen were fully determined. The pressure of other practical work has made it impossible to pursue the study with rapidity. It would be desirable to ascertain the effect of rate of motion over a greater range of speeds and the influence of air saturation; also to study other concentrations of dissolved oxygen. The data presented herein are sufficient to show that progress in corrosion study must be accomplished by thorough knowledge of these factors. ACKNOWLEDGXE NT The authors take pleasure in acknowledging the valuable advice and assistance received from E. Ward Tillotson, assistant director of Mellon Institute of Industrial Research, where the research was conducted by the junior author.

An Application of t h e Vacuum Tube to Chemistry’ By D. F. Calhane and R. E. Cushing WORCESTER POLYTECHNIC INSTITUTE; WORCESTER, MASS.

URING some work

D

FUNCTION OF THREE-ELECThe following article details a new method of accurately titrating involving the use of TRODE TUBE dilute solutions of salt with silver nitrate. The concentrations inthe w e 11- k n o w n vestigated varied from N/IOO to N/2500. I n the three-electrode Mohr reaction, it became The voltage changes produced in a concentration cell during titratube a filament of fine wire, necessary to employ this in tion of a solution in one limb of it are impressed on the grid element plain, or oxide-coated, is the titration of very dilute of the tube. heated to a high temperasolutions of salt, obtained The changes in potential of the grid oary the value of the plate ture by a battery current. as a discharge from the test current through the galvanometer, altering the magnitude of the scale The high vacuum inside the of a 3000-horse power turreading of the galvanometer. The end of the titration is shown by no tube allows negatively bine. As may be rememfurther change in the galoanometer deflection, when more titration charged particles or elecbered, this method employs reagent is added to the solution in one limb of the concentration cell. trons to be shot off from a solution of nitrate of the hot filament. These silver that is added to precipitate the chlorine in a neutral solution as silver chloride. electrons are drawn over to a positively charged plate, The end of the titration is indicated by a red color, due to the second element of the tube. The positive charge on the formation of silver chromate in presence of potassium the plate is produced by connecting it with the positive terminal of a second or plate battery. The negative terminal chromate as an indicator. Under favorable conditions this is a very accurate titration. of this plate battery is connected with the negative side of the I n the presence of any colored impurity in the water, or con- filament circuit. This provides a path for a current from the siderable precipitated silver chloride, the observance of the plate to the filament. If a galvanometer or milliammeter is red color, or end point, is a matter of considerable difficulty. connected into this circuit, a current will be indicated from It then becomes necessary in some cases to remove the color the plate to the filament; the value of this current, within certain ranges, will be dependent on the magnitude of the filaof the water by chemical means. It occurred to the writers that they,might employ the three- ment current that conditions the temperature of the Qament electrode tube, so widely used in radio work, as a means of and the consequent flow of electrons. If a third element, or grid, is placed between the filament indicating the end point, without the necessity of relying on and the plate, a means of controlling the magnitude of the any color change. It was found on consulting the literature that W. A. Noyes, plate current in two directions is available. If the grid is Jr.,2 had described a use of the three-electrode tube for the given a positive charge, it will attract negative electrons from measurement of potential in an electrolytic cell for the de- the filament. A current will also be established in the grid position of iron. This avoided the shunting of an appreciable circuit. If the grid is made negative, a reverse effect will be amount of current around the cell as would be the case when obtained and the plate current will be decreased for a given change in the potential of the grid. With a negative charge an ordinary voltmeter is used. A brief description of the manner in which a three-electrode the grid, located between plate and filament, repels the negaill make clearer the method of procedure in tive electrons from the filament back to the filament, and thus tube functions w reduces the value of the plate current in proportion to any this work. change that mzy occur in the value of the negative potential 1 Received January 27, 1923. of the grid. These changes in value of the plake current will 2 “Some Aspect of Electrolytic Iron,” Trans. Am. Electrochem. SOL, 40, be indicated by a galvanometer in the plate circuit. Ameans preprint (1921).